Wilcoxon test

Wilcoxon test should be associated with info about:

• Design (methods section, please address)

  • type of Wilcoxon test (i.e. Wilcoxon signed rank test, rank sum test (Mann-Whitney))

Wilcoxon signed rank test

• Assumptions to check: 

• sample of subjects has been randomly selected from the population it represents

• scores obtained for each of the subjects are interval/ratio data

• distribution of the difference scores in the population represented is symmetric

•  variables do not satisfy any distribution curve, even after transformation

• Report design  (in Methods section)

• one sample test or two sample matched pairs test

• one or two tails

• statistical package or program used (RRID)

• Report statistics (in Result section)

  • T-statistic (or z-statistic) exact p-values, group size

  • descriptive statistics

  • effect size

• Examples:

• Methods section:  The Wilcoxon test was used to evaluate the hypothesis. We used GraphPad Prism (RRID:SCR_002798) to perform the  analysis.

• Results section: A Wilcoxon test was performed to evaluate whether the reaction time for the group before treatment (n=35) had the same values as the reaction time for the group after treatment. The results indicated that the group after treatment  had significantly greater reaction time than the group before with  z = -2.05, p = .040 with medium effect size r=0.48. 

Wilcoxon rank sum test (Mann-Whitney)

• Assumptions to check

  • each sample has been randomly selected from the population it represents

  • two samples are independent of one another

  • data are at least ordinal

  • the populations distributions are identical in shape but does not have to be normal

  • Report design (in Methods section)

  • statistical package or program used (RRID)

  • Report statistics (in result section)

  • * U-statistic (or z-statistics for large groups), exact p-values, sample and group sizes * effect size * descriptive statistics